INDEX

Explanations

terms related to opposition or condemnation

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "the"

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

elaide

-0.80

herer

-0.77

 ILCS

-0.75

Leod

-0.69

ngth

-0.68

Iterator

-0.68

 survives

-0.67

lette

-0.67

river

-0.67

hd

-0.67

POSITIVE LOGITS

 idea

1.16

 latter

1.14

 notion

1.14

 slightest

1.12

 same

1.09

 possibility

1.08

 latest

1.06

 outcome

1.00

 prevailing

1.00

 imposition

0.97

Activations Density 0.692%

terms related to opposition or condemnation

occurrences of the word "the"

No Comments

No Known Activations